Comparison · Markets
How to compare mortgage markets without ranking places
Public mortgage records can describe where applications were reported and how their mix differed. They do not, by themselves, measure affordability, housing demand, or whether one place is a better market.
The short answer
Choose the geography that matches your question, align the cohort, and read counts beside shares. Treat HMDA as evidence of lender-reported mortgage activity—not a home-price index, affordability score, demand forecast, or personal outcome model.
01 · Define the job
A useful comparison begins with one answerable question
“Which market is best?” combines several different decisions. A focused question is more useful: Where was home-purchase application activity larger? How did refinance mix change? Did the share of reported actions move between two years?
It does not count cash purchases, rental activity, every inquiry for credit, or people who considered buying but never applied. A mortgage-application total is therefore not a complete measure of housing demand.
02 · Pick the unit
State, county, and metro areas answer different questions
A metro title names principal cities, but the statistical area generally includes whole component counties. Do not interpret a metro figure as a city-limits figure, and do not compare a county directly with a multi-county metro as if their scale were equivalent.
03 · Comparison contract
Hold the cohort constant before interpreting a difference
- The same activity year
- The same geographic level and boundary vintage
- The same application-action denominator
- The same loan purpose and product scope
- Counts shown beside shares
- Enough records to support a stable comparison
MortgagesVine's current state comparison uses 2024 HMDA actions 1, 2, 3, 4, 5, 7, and 8 and excludes purchased loans. Its year-over-year column applies the same definition to 2023. Changing that denominator changes the resulting shares.
Geographic definitions can also change. CFPB notes that multi-year HMDA comparisons can be limited by changes to metropolitan definitions, census-tract boundaries, field values, and reporting thresholds.
04 · Read the count
Application volume reflects market size and market mix
A more populous state will often report more applications than a smaller state. Housing stock, ownership, construction, refinancing conditions, lender coverage, and product mix can also move the count. Raw totals are useful for scale, but not as a score of market strength or consumer opportunity.
A year-over-year increase means more records appeared under the same cohort contract. It does not establish why the count changed, whether home prices rose, or whether individual households found borrowing more affordable.
06 · Complete the picture
Add context before making a housing-market claim
HMDA is a mortgage-lending dataset. Questions about prices, affordability, supply, taxes, or economic growth require additional sources and definitions.
- Population and household count
- Housing stock, tenure, and new construction
- Home prices, rents, income, and property taxes
- Mortgage rates and available loan products
- Local employment and migration
- Reporting coverage and boundary changes
Use outside measures with compatible dates and boundaries. Combining a current price series with an older mortgage cohort can create a comparison that looks precise but describes two different periods.
07 · Apply the method
Move from orientation to a defensible comparison
- Write the specific question and geographic level.
- Confirm the year, source release, and boundary definition.
- Align the denominator, loan purpose, and product scope.
- Compare counts first, then shares and year-over-year change.
- List plausible mix effects before explaining a difference.
- Add compatible housing and economic context if the claim requires it.
08 · Go to the record